Systems and methods for recommended sorting of search results for online searching

Abstract

A method may include receiving data related to a plurality of items and processing the data using a machine learning model. The machine learning model may have been trained to output a score for each of the plurality of items based on one or more target variables and to process the data using a grouped linear regression for groups of items based on sub-divisions of the groups. The method may include storing the output in a data store. Each entry in the data store may include at least an item identifier for an item, a group name, and the score. The method may include receiving search criteria for a search and identifying a set of search results in a group of items. The method may include determining an order of the set of search results and outputting the set of search results.

What is claimed is: 
     
         1 . A computer-implemented method for ordering search results, comprising:
 receiving item data related to a plurality of items;   applying a trained machine learning model to the item data to determine a plurality of scores corresponding to the plurality of items, wherein the plurality of items are grouped into a plurality of groups of items, and the trained machine learning model is applied on a group by group basis to the plurality of groups of items;   storing the plurality of scores corresponding to the plurality of items in a data store, wherein each score of the plurality of scores is stored in association with an identifier of one of the plurality of groups that the corresponding item from the plurality of items is grouped in;   receiving, from a user device, criteria for a search of the plurality of items;   identifying, from the data store, a group of items from the plurality of groups of items as a set of search results based on the criteria;   determining an order to present, within the set of search results, a subset of the plurality of items that are included in the group of items identified based on the score corresponding to each item in the subset of the plurality of items;   causing display of the set of search results in the order on the user device; and   receiving an indication of a user interaction with an item within the set of search results displayed on the user device, the user interaction indicative of a relevance of the item, wherein the trained machine learning model is retrained based on the user interaction to cause an adjustment to the score of the item determined using the retrained model and stored in the data store for use in future searches.   
     
     
         2 . The computer-implemented method of  claim 1 , wherein applying the trained machine learning model to the item data on a group by group basis comprises:
 processing first item data for a first group of items, of the plurality of groups of items, using a first processing operation; and   processing second item data for a second group of items, of the plurality of groups of items, using a second processing operation different from the first processing operation.   
     
     
         3 . The computer-implemented method of  claim 1 , wherein the item data includes a plurality of target variable values for the plurality of items, and applying the trained machine learning model to the item data comprises:
 determining the plurality of scores corresponding to the plurality of items based on the plurality of target variable values for the plurality of items.   
     
     
         4 . The computer-implemented method of  claim 3 , wherein the trained machine learning model is further configured to process a group of items, of the plurality of groups of items, based on sub-divisions of the group of items. 
     
     
         5 . The computer-implemented method of  claim 4 , wherein the sub-divisions comprise pentiles or deciles for a subset of the plurality of target variable values within the group of items. 
     
     
         6 . The computer-implemented method of  claim 1 , wherein the item data defines a plurality of item variables associated with the plurality of items, and grouping the plurality of items into the plurality of groups of items comprises:
 grouping the plurality of items according to one or more of the plurality of item variables.   
     
     
         7 . The computer-implemented method of  claim 1 , wherein the plurality of scores corresponding to the plurality of items are determined relative to a target variable, and the method further comprises:
 performing a path analysis of multiple target variables; and   selecting the target variable, from the multiple target variables, based on a result of the path analysis.   
     
     
         8 . The computer-implemented method of  claim 7 , wherein the target variable includes views or user selections of an item details page for each of the plurality of items. 
     
     
         9 . The computer-implemented method of  claim 1 , further comprising:
 receiving a user profile of a user associated with the user device; and   providing the user profile to the trained machine learning model, wherein the trained machine learning model is applied to the item data and the user profile to determine the plurality of scores corresponding to the plurality of items, wherein the plurality of scores are customized for the user profile.   
     
     
         10 . The computer-implemented method of  claim 9 , wherein the trained machine learning model is a first machine learning model, and the method further comprises:
 receiving user data of the user associated with the user device; and   applying, to the user data, a second machine learning model trained to generate and output the user profile.   
     
     
         11 . The computer-implemented method of  claim 10 , wherein the user data includes one or more of a location of the user, previous search criteria associated with the user, or user interactions with previous sets of search results. 
     
     
         12 . The computer-implemented method of  claim 1 , wherein the user interaction includes a selection of the item within the set of search results indicative of a higher relevance of the selected item than non-selected items. 
     
     
         13 . The computer-implemented method of  claim 12 , wherein the trained machine learning model is retrained based on the selection to one or more of:
 reduce the score of the non-selected items or increase the score of the selected item.   
     
     
         14 . A computer-implemented method for ordering vehicle search results, comprising:
 receiving vehicle data related to a plurality of vehicles;   applying a trained machine learning model to the vehicle data to determine a plurality of scores corresponding to the plurality of vehicles, wherein the plurality of vehicles are grouped into a plurality of vehicle groups, and the trained machine learning model is applied on a group by group basis to the plurality of vehicle groups;   storing the plurality of scores corresponding to the plurality of vehicles in a data store, wherein each score of the plurality of scores is stored in association with an identifier of one of the plurality of vehicle groups that the corresponding vehicle from the plurality of vehicles is grouped in;   receiving, from a user device, criteria for a vehicle search;   identifying, from the data store, a vehicle group from the plurality of vehicle groups as a set of search results based on the criteria;   determining an order to present, within the set of search results, a subset of the plurality of vehicles that are included in the vehicle group identified based on the score corresponding to each vehicle in the subset;   causing display of the set of search results in the order on the user device; and   receiving an indication of a user interaction with a vehicle within the set of search results displayed on the user device, the user interaction indicative of a relevance of the vehicle, wherein the trained machine learning model is retrained based on the user interaction to cause an adjustment to the score of the vehicle determined using the retrained model and stored in the data store for use in future searches.   
     
     
         15 . The computer-implemented method of  claim 14 , wherein the vehicle data includes a plurality of target variable values for the plurality of vehicles, and applying the trained machine learning model to the vehicle data comprises:
 determining the plurality of scores corresponding to the plurality of vehicles based on the plurality of target variable values for the plurality of vehicles.   
     
     
         16 . The computer-implemented method of  claim 15 , wherein applying the trained machine learning model to the vehicle data on the group by group basis comprises:
 for a vehicle group, of the plurality of vehicle groups, including a subset of the plurality of vehicles, processing the vehicle data for the subset of the plurality of vehicles based on sub-divisions of the vehicle group.   
     
     
         17 . The computer-implemented method of  claim 14 , wherein grouping the plurality of vehicles into the plurality of vehicle groups comprises:
 grouping the plurality of vehicles into the plurality of vehicle groups based on one or more of a vehicle make, a vehicle model, a vehicle production year, a vehicle body style, or a vehicle condition defined by the vehicle data.   
     
     
         18 . The computer-implemented method of  claim 14 , further comprising:
 receiving a user profile of a user associated with the user device; and   providing the user profile to the trained machine learning model, wherein the trained machine learning model is applied to the vehicle data and the user profile to determine the plurality of scores corresponding to the plurality of vehicles, wherein the plurality of scores are customized for the user profile.   
     
     
         19 . The computer-implemented method of  claim 18 , wherein the user profile is based one or more of a location of the user, previous search criteria associated with the user, or user interactions with previous sets of search results. 
     
     
         20 . A computer-implemented method for training a machine learning model to facilitate ordering search results, comprising:
 receiving a plurality of training datasets for a plurality of reference items, each training dataset of the plurality of training datasets including reference item data, a target variable value, and a ground truth score;   grouping the plurality of reference items into a plurality of item groups based on reference item data;   training a machine learning model, based on the plurality of training datasets, to determine, on a group by group basis, a score for a reference item within one of the plurality of item groups based on the target variable value and the ground truth score, wherein, upon a completion of the training, the trained machine learning model is configured to determine a plurality of scores corresponding to a plurality of items based on received item data for the plurality of items, including target variable values for the plurality of items, that are stored in a data store for use in searches for the plurality of items;   receiving an indication of a user interaction associated with an item from the plurality of items having been presented within a set of search results to a user device based on search criteria received from the user device and a score of the item stored in the data store; and   retraining the trained machine learning model based on the user interaction to cause an adjustment to the score of the item, where the adjusted score for the item is stored in the data store for use in future searches.